e000a3bb0d
fix(non-interactive-env): force unix prefix for bash git commands
...
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent )
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai >
2026-03-25 01:23:02 +09:00
0b77e2def0
Merge pull request #2810 from code-yeongyu/fix/webfetch-redirect-loop
...
fix(webfetch): guard redirect loops in built-in flow
2026-03-25 00:40:54 +09:00
d327334ded
Merge pull request #2808 from code-yeongyu/fix-gemini-3-pro-cleanup
...
fix(models): remove stale Gemini 3 Pro references
2026-03-25 00:32:10 +09:00
971912e065
fix(webfetch): avoid rewriting successful redirect content
2026-03-24 23:59:57 +09:00
af301ab29a
fix(webfetch): guard redirect loops in built-in flow
2026-03-24 23:58:53 +09:00
535ecee318
fix(models): remove stale Gemini 3 Pro references
...
Keep repo-owned CLI, docs, and test fixtures aligned with current Gemini 3.1 naming while leaving upstream catalog behavior untouched.
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent )
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai >
2026-03-24 23:53:56 +09:00
a0649616bf
fix(todo-continuation-enforcer): acknowledge compaction epochs during idle
...
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent )
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai >
2026-03-24 20:36:22 +09:00
cb12b286c8
fix(todo-continuation-enforcer): arm compaction epochs on compaction
...
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent )
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai >
2026-03-24 20:36:22 +09:00
8e239e134c
fix(todo-continuation-enforcer): make compaction guard epoch-aware
...
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent )
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai >
2026-03-24 20:36:22 +09:00
733676f1a9
fix(todo-continuation-enforcer): add compaction epoch state
...
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent )
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai >
2026-03-24 20:36:22 +09:00
d2e566ba9d
fix(preemptive-compaction): mock session history in degradation test
...
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent )
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai >
2026-03-24 20:36:22 +09:00
0e7bd595f8
fix(session-recovery): reuse signed thinking blocks safely
...
Reuse signed Anthropic thinking blocks only when they can still sort before the target message's parts, otherwise skip recovery instead of reintroducing invalid loops.
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent )
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai >
2026-03-24 17:22:07 +09:00
0732cb85f9
fix(thinking-block-validator): reuse signed thinking parts
...
Preserve prior signed Anthropic thinking blocks instead of creating unsigned synthetic placeholders, and skip injection when no signed block exists.
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent )
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai >
2026-03-24 17:22:07 +09:00
69d0b23ab6
fix(anthropic-recovery): clear session state after successful summarize and fix timing test
...
- Add missing clearSessionState() call after successful summarize (line 117)
Without this, retry state persisted even after success, potentially causing
unnecessary retries on subsequent compaction events.
- Fix timing-sensitive test: adjust attempt=0 and firstAttemptTime to give
proper remainingTimeMs buffer for capped delay calculation.
Fixes #2225
Supersedes #2671
2026-03-24 16:23:11 +09:00
935995d270
Merge pull request #2668 from MoerAI/fix/session-degradation-detection
...
fix(session): detect post-compaction no-text degradation and trigger recovery (fixes #2232 )
2026-03-24 16:21:30 +09:00
23d8b88c4a
Merge pull request #2669 from MoerAI/fix/atlas-worktree-verification
...
fix(atlas): use worktree path for git verification when available (fixes #2229 )
2026-03-24 16:21:27 +09:00
f1eaa7bf9b
fix(shell): detect csh/tcsh and use setenv syntax ( #2769 )
...
fix(non-interactive-env): detect shell type for csh/tcsh env var syntax (fixes #2089 )
2026-03-24 12:30:49 +09:00
230ce835e5
fix: resolve 3 bugs - subagent model override, empty plan completion, deep task refusal
...
- #2741 : Pass inheritedModel as fallback in subagent-resolver when user hasn't
configured an override, ensuring custom provider models take priority
- #2648 : Fix getPlanProgress to treat plans with 0 checkboxes as incomplete
instead of complete (total > 0 && completed === total)
- #2779 : Relax Hephaestus single-task guard to accept multi-step sub-tasks
from Atlas delegation, only rejecting genuinely independent tasks
Fixes #2741 , fixes #2648 , fixes #2779
2026-03-24 09:45:11 +09:00
aaaeb6997c
fix(ralph-loop): add semantic completion detection as fallback for natural language ( fixes #2489 )
2026-03-23 20:46:10 +09:00
2f801f6c28
fix(hooks): add bash-file-read-guard to warn agents against cat/head/tail ( fixes #2096 )
2026-03-23 20:15:20 +09:00
92509d8cfb
fix(non-interactive-env): detect shell type for csh/tcsh env var syntax ( fixes #2089 )
2026-03-23 19:33:54 +09:00
94c234c88c
Merge pull request #2765 from code-yeongyu/fix/issue-2024
...
fix: skip keyword injection for non-OMO agents (fixes #2024 )
2026-03-23 18:39:48 +09:00
de371be236
fix: skip keyword injection for non-OMO agents ( fixes #2024 )
2026-03-23 18:10:44 +09:00
a64e364fa6
fix: clarify Prometheus file permission error message ( fixes #2150 )
2026-03-23 18:07:59 +09:00
f16d55ad95
fix: add errorName-based quota detection and strengthen test coverage
2026-03-23 15:19:09 +09:00
62d2704009
fix(runtime-fallback): detect prettified quota errors without HTTP status codes ( fixes #2747 )
2026-03-23 10:34:22 +09:00
5777bf9894
fix(atlas): restore agent mismatch guard for subagent boulder continuation ( #18681 )
2026-03-23 01:04:36 +00:00
07ea8debdc
fix(keyword-detector): remove duplicate separator from ultrawork templates
2026-03-21 19:09:51 +09:00
10eb3a07e0
Merge branch 'dev' into fix/analyze-mode-load-skills-hint
2026-03-20 16:29:13 +09:00
03feaa0594
fix(analyze-mode): add mandatory load_skills param hint to prevent delegate_task errors
2026-03-20 16:26:17 +09:00
8aec4c5cb3
feat(hooks/todo-continuation-enforcer): enhance continuation message with skeptical verification guidance
2026-03-20 16:13:02 +09:00
3773e370ec
fix(runtime-fallback): detect bare 429 rate-limit signals ( fixes #2677 )
2026-03-20 11:00:00 +09:00
0e610a72bc
fix(runtime-fallback): make fallback provider selection provider-agnostic ( fixes #2303 )
2026-03-20 09:53:24 +09:00
3aa2748c04
docs: sync hook counts after continuation hook removal
...
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent )
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai >
2026-03-19 12:02:52 +09:00
ccaf759b6b
fix(hooks): remove gpt permission continuation hook
...
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent )
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai >
2026-03-19 12:02:52 +09:00
521a1f76a9
fix(atlas): stop only after 10 consecutive prompt failures
...
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent )
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai >
2026-03-19 12:02:52 +09:00
1f64a45113
Merge pull request #2620 from code-yeongyu/feat/openclaw-bidirectional
...
feat: port OpenClaw bidirectional integration from omx
2026-03-19 10:47:07 +09:00
9b2dc2189c
fix(ralph-loop): detect promise tags in tool_result parts for ulw verification
...
Oracle's <promise>VERIFIED</promise> arrives as a tool_result part from the
task() tool call, not as a text part. Both detectCompletionInSessionMessages
and collectAssistantText only scanned type=text parts, missing the
verification signal entirely. This caused ulw loops to fail verification
even when Oracle successfully emitted VERIFIED.
Include tool_result parts in promise detection alongside text parts.
Exclude tool_use parts to avoid false positives from instructional text.
2026-03-18 19:09:59 +09:00
071fab1618
fix: match existing codebase session.messages() parameter shape
2026-03-18 19:08:05 +09:00
f6c24e42af
fix(ralph-loop): detect promise tags in tool_result parts for ulw verification
...
Oracle's <promise>VERIFIED</promise> arrives as a tool_result part from the
task() tool call, not as a text part. Both detectCompletionInSessionMessages
and collectAssistantText only scanned type=text parts, missing the
verification signal entirely. This caused ulw loops to fail verification
even when Oracle successfully emitted VERIFIED.
Include tool_result parts in promise detection alongside text parts.
Exclude tool_use parts to avoid false positives from instructional text.
2026-03-18 19:03:30 +09:00
22fd976eb9
feat(categories): change quick category default model from claude-haiku-4-5 to gpt-5.4-mini
...
GPT-5.4-mini provides stronger reasoning at comparable speed and cost.
Haiku remains as the next fallback priority in the chain.
Changes:
- DEFAULT_CATEGORIES quick model: anthropic/claude-haiku-4-5 → openai/gpt-5.4-mini
- Fallback chain: gpt-5.4-mini → haiku → gemini-3-flash → minimax-m2.5 → gpt-5-nano
- OpenAI-only catalog: quick uses gpt-5.4-mini directly
- Think-mode: add gpt-5-4-mini and gpt-5-4-nano high variants
- Update all documentation references
2026-03-18 19:03:30 +09:00
3c7e6a3940
fix(atlas): address review findings for task session reuse
2026-03-18 18:44:42 +09:00
33ef4db502
Merge pull request #2640 from HaD0Yun/had0yun/atlas-task-session-reuse
...
feat(atlas): persist preferred task session reuse
2026-03-18 18:37:16 +09:00
458ec06b0e
fix: extract question text from questions array per opencode tool schema
2026-03-18 18:27:09 +09:00
6b66f69433
feat(gpt-permission-continuation): add context-aware continuation prompts
...
- Add buildContextualContinuationPrompt to include assistant message context
- Move extractPermissionPhrase to detector module for better separation
- Block continuation injection in subagent sessions
- Update handler to use contextual prompts with last response context
- Add tests for subagent session blocking and contextual prompts
- Update todo coordination test to verify new prompt format
🤖 Generated with assistance of OhMyOpenCode
2026-03-18 17:52:32 +09:00
ce8957e1e1
fix(ralph-loop): harden oracle verification flow
...
Capture oracle verification sessions more reliably and accept parent-session VERIFIED evidence so ULW loops do not retry after successful review.
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent )
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai >
2026-03-18 17:45:59 +09:00
a3db64b931
fix: address cubic review — SDK compatibility and race condition fixes
2026-03-18 17:42:17 +09:00
8859da5fef
fix(atlas): harden task session reuse
2026-03-18 17:31:27 +09:00
4723319eef
fix(atlas): use worktree path for git verification when available ( fixes #2229 )
2026-03-18 16:23:37 +09:00
b8f3186d65
fix(session): detect post-compaction no-text degradation and trigger recovery ( fixes #2232 )
2026-03-18 16:13:23 +09:00
YeonGyu-Kim